search

dpservis / bots

Automatically exported from code.google.com/p/bots
0 stars 0 forks source link

new: extend chained translations #136

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago 
option to use output of one translation as input of another.

this is done by returning an dict from an translation script;
'type' indicates the type of chained translation, 
'alt' indicates the alt value to be used.

eg in mapping script:
    return {'type':'out_as_inn','alt':'test'}

Original issue reported on code.google.com by hjebb...@gmail.com on 10 May 2012 at 5:39

GoogleCodeExporter commented 8 years ago 
Hi henk-jan,
Testing this... slightly strange results.

If my script just has
    return {'type':'out_as_inn','alt':'EMAIL'} # convert this mapped idoc to an email

Then result is 1 file in, 2 files out. But the second translation is missing 
all it's data.

If my script has
    out.ta_info['statust'] = DONE # discard this idoc
    return {'type':'out_as_inn','alt':'EMAIL'} # convert this mapped idoc to an email

Then result is ok, 1 file out (the second translation) and all the data is ok. 
However I cannot view the "intermediate" format anywhere in view details. This 
is not totally a problem but would be nice to be able to see it.

If I wanted both outputs (I don't in this case but someone might) then this 
does not work properly. I have attached a couple of screenshots.

Original comment by mjg1964 on 11 May 2012 at 1:17

Attachments:

GoogleCodeExporter commented 8 years ago 
With my existing routescript method it looks like this. I can view the 
intermediate format.

Original comment by mjg1964 on 11 May 2012 at 1:22

Attachments:

GoogleCodeExporter commented 8 years ago 
hi Mike,

see plugin for setup that tests this.

Original comment by hjebb...@gmail.com on 11 May 2012 at 1:36

GoogleCodeExporter commented 8 years ago 
I can't see a plugin for this, where do I find it?

Original comment by mjg1964 on 11 May 2012 at 11:25

GoogleCodeExporter commented 8 years ago 
hi mike,

I attached in the previous mail (which I also send to 
mike.griffin@clipsal.com.au

Original comment by hjebb...@gmail.com on 11 May 2012 at 11:37

GoogleCodeExporter commented 8 years ago 
hi Mike,

have you been able to test this?

Original comment by hjebb...@gmail.com on 15 May 2012 at 5:50

GoogleCodeExporter commented 8 years ago 
Hi henk-jan. I have just been testing this some more. Your plugin works ok, but 
your alt mapping just does inn2out. Mine is idoc to html-template, my inn.get 
do not get any data from the input! will do more testing...

Original comment by mjg1964 on 16 May 2012 at 7:43

GoogleCodeExporter commented 8 years ago 
hai Mike,

my plugin is  my test for the type of translation, it is somewaht more complex
what it does:
- translate order to fixed
- translate order to aperak (alt-translation)
- translate aperak to aperak
so: 1 incoming order generates:
- one fixed order
- 2 aperaks

(sorry, should have made mentioned this earlier)

Original comment by hjebb...@gmail.com on 16 May 2012 at 7:48

GoogleCodeExporter commented 8 years ago 
yes, I understand your plugin and got it to work ok. The part that interests me 
is the last step.
I want to:
- translate edifact to idoc order
if (some conditions):
  - translate the idoc to html-template (alt-translation)
  - discard the idoc

Since yours works, I need to find out why mine doesn't. In my alt mapping 
script, inn.root.display shows the idoc data but inn.get's do not retrieve any 
data. However using my original routescript method this mapping does work ok.

Original comment by mjg1964 on 16 May 2012 at 8:39

GoogleCodeExporter commented 8 years ago 
A clue! I tried an inn2out translation with the idoc, it fails (note the 
trailing spaces).
MessageError: Grammar 
"D:\python27\lib\site-packages\bots\bots.usersys\grammars\idoc.ORDERS05" has 
(root)record "EDI_DC40"; found "EDI_DC40 ".

Original comment by mjg1964 on 16 May 2012 at 11:16

GoogleCodeExporter commented 8 years ago 
So I have to do my inn.get like this! (spaces replaced by dots for clarity)

inn.get({'BOTSID':'EDI_DC40...'},{'BOTSID':'E1EDK01.......................'},{'B
OTSID':'E1EDK02.......................','QUALF':'001','BELNR':None})

Original comment by mjg1964 on 16 May 2012 at 11:44

GoogleCodeExporter commented 8 years ago 
hi Mike,

I am not sure what you found out.
is this working for you?

Original comment by hjebb...@gmail.com on 17 May 2012 at 3:01

GoogleCodeExporter commented 8 years ago 
Hi henk-jan,
It does not work "correctly" for idoc (and probably fixed) editypes, because 
the tomessage from first translation contains all the padding spaces according 
to the field lengths. So when that tomessage is used as inn for the next 
translation (transform.py line 125), all the get's must specify the correct 
number of padding spaces otherwise they return None. This is different to a 
normal mapping which of course does not require this.

eg. in the 'out_as_inn' mapping:
inn.get({'BOTSID':'EDI_DC40'},{'BOTSID':'E1EDK01'} does not work
inn.get({'BOTSID':'EDI_DC40...'},,{'BOTSID':'E1EDK01.......................'} 
works

Original comment by mjg1964 on 17 May 2012 at 10:31

GoogleCodeExporter commented 8 years ago 
I guess this is because the BOTSID field is longer than the actual identifier,
bots writes the original outgoing file, and during writing the fields lengts 
are adjusted.

any suggestions on how to solve this?

can think of:
1. monkey-patch it: if fixed (idoc is a subclass of fixed I think) than run 
over tree and get rid of spaces
2. write a copy of the tree; mmmmh, copying is quite slow.....

Original comment by hjebb...@gmail.com on 17 May 2012 at 10:52

GoogleCodeExporter commented 8 years ago 
I did try a couple of things to fix it, but I don't know much about this part 
of the code. Removing the spaces with a function in transform.py probably the 
easiest way.

Original comment by mjg1964 on 17 May 2012 at 11:47

GoogleCodeExporter commented 8 years ago 
and it is not only BOTSID, the same issue applies to all fields used in inn.get

Original comment by mjg1964 on 17 May 2012 at 11:54

GoogleCodeExporter commented 8 years ago 
yes, I think so.
all fields are formatted as for output.
probably empty fields are added, to.

Original comment by hjebb...@gmail.com on 18 May 2012 at 12:04

GoogleCodeExporter commented 8 years ago 
wouldn't be to hard:
def stripnode(node):
     if node.record is not None:
         node.record['BOTSID'] = node.record['BOTSID'].strip()
     for child in node.children:
         stripnode(child)

Original comment by hjebb...@gmail.com on 18 May 2012 at 12:09

GoogleCodeExporter commented 8 years ago 
Thabks! I got it working but will do a little more testing.

transform.py at line 125
    inn = tomessage
    stripnode(inn.root)

At line 153 added this function (added another loop)
def stripnode(node):
    if node.record is not None:
        for field in node.record:
            node.record[field] = node.record[field].strip()
    for child in node.children:
        stripnode(child)

Original comment by mjg1964 on 18 May 2012 at 3:05

GoogleCodeExporter commented 8 years ago 
Changed this so stripnode is only done for fixed formats.

    inn = tomessage
    if isinstance(inn,outmessage.fixed):
        stripnode(inn.root)

This now works perfectly for me. The only thing missing, but not really 
important; if I discard the intermediate format idoc (out.ta_info['statust'] = 
DONE) then I can't go back and view it later in bots monitor.

Original comment by mjg1964 on 18 May 2012 at 4:58

Attachments:

GoogleCodeExporter commented 8 years ago 
ok, will add this.
....looking at this again, looks like stripnode() is a method in class Node.

Original comment by hjebb...@gmail.com on 18 May 2012 at 11:28

GoogleCodeExporter commented 8 years ago

Original comment by hjebb...@gmail.com on 10 Sep 2013 at 12:45